three-dimensional child...
TRANSCRIPT
Three-Dimensional Child Anthropometry for Vehicle Safety Analysis
Matthew P. Reed Sheila M. Ebert-Hamilton
Biosciences Group October 2012
Children in Cars Data on the size, shape, and posture of children are needed for:
CRS Design!
Optimizing Belt Restraints!
Developing Computational Models!
Developing ATDs!
Child Anthropometry • The most recent large-scale, detailed
study of U.S. children was conducted by UMTRI (HSRI) in the 1970s for the US Consumer Product Safety Commission
• The ongoing U.S. NHANES gathers stature, body weight, and a few other dimensions, but this information is insufficient for product design and analysis
Standard Anthropometry Anthropometers, calipers, and tape measures: 1D dimensions
Snyder et al. (1977)!
Child Body Shells The current Hybrid-III 3YO and 6YO ATDs are based on standard anthropometry and 3-D surface representations based on 1-D data created in the early 1970s
3YO!
6YO!
Young et al. (1975)!
Functional Anthropometry Measure physical attributes in task-relevant conditions
Measuring belt fit Measuring posture by digitizing the 3D locations of body landmarks
Multivariate Functional Anthro
Measure Landmarks in Seated Postures!
Principal Component
Analysis!
Vehicle and Seat Geometry!
Joint Center Location Estimates! Crash Dummy Anthro Specs!
Regression!
Target Body Dimensions: Stature, Body Weight, …!
Large Child Omnidirectional Dummy
(Humanetics ATD)
Whole-Body Landmark & Joint Configurations!
UMTRI Child Body Shape Study Objective: Quantify body shape and vehicle seating postures for children
ages 4 to 11
Posture and belt fit in vehicle seating
Standard anthropometry
Whole-body scanning
Subject Pool Total 162 children (78 boys, 82 girls), ages 4 to 11
Hybrid III Reference Dimensions
3YO 6YO
10YO
Small Female Adult
Methods – Standard Anthro 23 standard anthropometric dimensions taken to document child size
Methods match previous large-scale UMTRI child anthropometry study (Snyder et al. 1977) where possible
Methods – Landmarking • Total of 92 landmarks
measured directly (FARO Arm) or digitized in scan data
• 33 landmarks measured with FARO Arm in vehicle seat and booster conditions
Methods – Vehicle Seat Child posture and belt fit measured in a midrange vehicle seating condition with and without a belt-positioning booster
Provides a direct linkage to previous UMTRI laboratory and in-vehicle child posture and belt-fit studies
Methods – Scanning Lasers travel top to bottom, “painting” a red line on the subject
Two cameras on each tower view laser line, convert to coordinates
Custom platforms for standing and seated conditions
VITUS XXL scanner from Human Solutions
12-second scan time, approximately 500k points per scan, depending on subject size
Four laser towers with eye-safe red light
Methods – Scanning
Pelvis points recorded in some postures using FARO Arm
Lap area (shadowed from Vitus scanner) was manually scanned using FARO-Arm laser scanner in some postures
Body Shape Modeling Whole-Body Scan Data
Clean and Fit Polygon Mesh
Manual Landmark Extraction
Manually Measured Body Landmarks
Handheld Scanner Data
Standard Anthropometry
Mesh with Landmarks Segment and Resample Mesh
Model Integration
PCA+Regression Analysis
Statistical Model to Predict Body Shape from Standard Anthro or Landmark Locations
Skeletal Anthropometry Parametric Modeling of Skeletal Structures
CT Images
Data Extraction (N > 100)
Statistical Modeling
Predicted Skeleton Size and Shape = ƒ (stature, mass, …)
2nd PC of Ribcage Shape
Hybrid-III vs. Human Pelvis
Model Development
Vehicle Environment • seat back angle • seat cushion angle • seat cushion length • booster/no-booster
Posture Prediction:
Landmark and joint locations relative to seat Child Attributes
• stature • body weight • gender?
Skeletal Geometry Prediction:
Bone meshes
Body Shape Prediction:
Surface mesh
Lab and In-Vehicle Child Posture Studies
Integrated Child Anthropometry
Model
Skeletal Modeling from CT
Body Shapes from Current Study
Next Steps and Future Work • Complete data analysis and modeling
• Develop body shape targets for crash test dummies across the range of relevant sizes and postures
• Apply results to morphing to human-body FE models
Acknowledgements
This research was funded by NHTSA under contract DTNH22-10-H-00288 with the University of Michigan!
!
!
Contacts: !
mreed.umtri.umich.edu!
!
Scan Data Processing First step is manually stripping props
Raw scan data for some of the study postures!
Scan Data Processing Visible landmarks manually extracted using Meshlab software from scans with grayscale texture
Multiple trials to quantify repeatability and reproducibility
Combined with FARO Arm coordinate measurements